ARTICLE

Toll-like receptor 2 deficiency improves insulin sensitivityand hepatic insulin signalling in the mouse

L.-H. Kuo & P.-J. Tsai & M.-J. Jiang & Y.-L. Chuang &

L. Yu & K.-T. A. Lai & Y.-S. Tsai

Received: 5 August 2010 /Accepted: 1 September 2010 /Published online: 22 October 2010# Springer-Verlag 2010

AbstractAims/hypothesis Substantial evidence suggests a link be-tween elevated inflammation and development of insulinresistance. Toll-like receptor 2 (TLR2) recognises a largenumber of lipid-containing molecules and transducesinflammatory signalling in a variety of cell types, includinginsulin-responsive cells. Considering the contribution of thefatty acid composition in TLR2-depedent signalling, wehypothesised that the inflammatory signals transduced byTLR2 contribute to insulin resistance.

Methods Mice deficient in TLR2 were used to investigatethe in vivo roles of TLR2 in initiating and maintaininginflammation-associated insulin resistance and energyhomeostasis.Results We first recapitulated the observation with elevatedexpression of TLR2 and inflammatory cytokines in whiteadipose tissue and liver of ob/ob mice. Aged or high-fat-fedTLR2-deficient mice were protected from obesity andadipocyte hypertrophy compared with wild-type mice.Moreover, mice lacking TLR2 exhibited improved glucosetolerance and insulin sensitivity regardless of feeding themregular chow or a high-fat diet. This is accompanied byreductions in expression of inflammatory cytokines andactivation of extracellular signal-regulated kinase (ERK) ina liver-specific manner. The attenuated hepatic inflamma-tory cytokine expression and related signalling are corre-lated with increased insulin action specifically in the liverin TLR2-deficient mice, reflected by increased insulin-stimulated protein kinase B (Akt) phosphorylation andIRS1 tyrosine phosphorylation and increased insulin-suppressed hepatocyte glucose production.Conclusions/interpretation The absence of TLR2 attenu-ates local inflammatory cytokine expression and relatedsignalling and increases insulin action specifically in theliver. Thus, our work has identified TLR2 as a key mediatorof hepatic inflammation-related signalling and insulinresistance.

Keywords Extracellular signal-regulated kinase .

Inflammation . Insulin resistance . Toll-like receptor 2

AbbreviationsAkt Protein kinase BBAT Brown adipose tissueERK Extracellular signal-regulated kinase

Electronic supplementary material The online version of this article(doi:10.1007/s00125-010-1931-5) contains supplementary material,which is available to authorised users.

L.-H. Kuo :Y.-S. Tsai (*)Institute of Basic Medical Sciences, College of Medicine,National Cheng Kung University,Tainan, Taiwan, Republic of Chinae-mail: yaustsai@mail.ncku.edu.tw

P.-J. TsaiDepartment of Medical Laboratory Science and Biotechnology,College of Medicine, National Cheng Kung University,Tainan, Taiwan, Republic of China

M.-J. Jiang :Y.-L. Chuang :Y.-S. TsaiInstitute of Clinical Medicine and Cardiovascular ResearchCenter, College of Medicine and Hospital,National Cheng Kung University,Tainan, Taiwan, Republic of China

L. YuInstitute of Behavioral Medicine, College of Medicine,National Cheng Kung University,Tainan, Taiwan, Republic of China

K.-T. A. LaiNovoTaiwan Biotech,Taipei, Taiwan, Republic of China

Diabetologia (2011) 54:168–179DOI 10.1007/s00125-010-1931-5

HF High fatIKK IκB kinaseITT Insulin tolerance testJNK c-Jun N-terminal kinaseMCP-1 Monocyte chemotactic protein-1NF-κB Nuclear factor kappa BPI3K Phosphoinositide 3-kinaseRC Regular chowTLR Toll-like receptorWAT White adipose tissue

Introduction

Insulin resistance is the chief abnormality present inmetabolic syndrome. Increasing evidence suggests causa-tive links between inflammation and development ofinsulin resistance [1, 2]. Thus, the inflammatory nuclearfactor kappa B (NF-κB) signalling pathway is activated andpro-inflammatory cytokines are produced in the whiteadipose tissue (WAT) and livers of insulin-resistant animalsand humans [3–5]. Genetic or pharmacological inhibition ofNF-κB pathways and cytokine neutralisation reversesinsulin resistance [4, 6]. Previous studies demonstrated thatinflammation-induced activation of protein kinases, such asIκB kinases (IKKs), c-jun N-terminal kinases (JNKs) andextracellular signal-regulated kinases (ERKs), inhibitstyrosine phosphorylation of IRS1 and suppresses down-stream insulin signalling. This results in inefficient glucoseuptake and use [7–9]. Thus, it is critical to identify thetarget(s) and mechanism(s) that initiate inflammatoryresponses leading to insulin resistance.

Insulin resistance is frequently associated with obesity[10] and dyslipidaemia [11]. Obesity is thought to be themost important identified factor contributing to insulinresistance [12]. Dyslipidaemia, characterised by abnormalamounts and composition of lipids and/or lipoproteins inblood, represents an additional risk factor for insulinresistance. NEFAs and triacylglycerols are elevated inobese and diabetic individuals and animals. Previousstudies demonstrated that exposure to elevated NEFAcauses activation of inflammatory signals and productionof cytokines in multiple cell types, and these findings havebeen correlated with impairments of insulin actions [13,14]. Thus, circulating lipid and associated inflammationappear to be a link between nutrient excess and insulinresistance. Recent studies have suggested that the inductionof inflammatory pathways by dietary factors may mediatethrough the membrane receptors, including toll-like recep-tors (TLRs) [15].

TLRs play a central role in the innate immuneresponse by recognising conserved patterns in diverse

pathogenic molecules and activating pro-inflammatorysignalling pathways [16, 17]. TLR2, a subclass of TLRs,is the main receptor recognising components of bacterialcell wall, lipoproteins and lipopeptides. Ligand-induceddimerisation of TLR2 with either TLR1 or TLR6 triggersrecruitment of adaptor proteins, following a cascade ofkinase activation, ultimately leading to activation of NF-κB and production of pro-inflammatory cytokines. Thebinding specificity of TLR2 with lipoprotein ligand ismediated through its dimerisation with either TLR1 orTLR6. The TLR2–TLR1 complex recognises triacylatedlipoprotein, whereas the TLR2–TLR6 complex sensesdiacylated lipoprotein [18]. Thus, the number of acylchain of lipoproteins is finely differentiated by TLR1–TLR2 and TLR2–TLR6. In addition to the number of acylchains, other characteristics, such as length and ester bondof acyl chains, are critical for the biological activity ofTLR2-dependent signalling [19, 20]. These results suggestthat the fatty acid composition plays a central role inligand recognition and receptor activation for TLR2.Furthermore, other studies demonstrated that TLR2 isinvolved in NEFA-induced insulin resistance [21–23]. It isreasonable to postulate that the fatty acid moiety fromnutrients could potentially activate TLR2 and transducethe inflammatory signals.

Although the predominant site of TLR2 production is oncells of the innate immune system [17], TLR2 is found on anumber of insulin-responsive cells, including adipose,skeletal muscle and liver cells [24, 25]. However, the rolesof TLR2 in initiating and maintaining inflammatory-associated insulin resistance and energy homeostasis invivo have not been established. Here, we show that micewith TLR2 deficiency are protected from developinginsulin resistance and adiposity. The increased insulinsensitivity is associated with increased insulin-signaltransduction, improved glucose metabolism, and reducedinflammatory cytokine expression and related signallingspecifically in the liver.

Methods

Mice Mice deficient in TLR2 (Tlr2−/−) were kindlyprovided by S. Akira [26] and maintained on a C57BL/6genetic background. Studies were carried out using bothmale and female Tlr2−/− mice and age-matched wild-type(WT) C57BL/6 mice, obtained from National Cheng KungUniversity Laboratory Animal Center. Leptin-deficient (ob/ob) mice were obtained from the Jackson Laboratory (BarHarbor, ME, USA). Mice were fed ad libitum either regularchow (RC) (Purina Laboratory Rodent Diet 5001, PMINutrition International, Richmond, IN, USA) or a high-fat(HF) diet (58Y1; TestDiet, Richmond, IN, USA).

Diabetologia (2011) 54:168–179 169

Tissue collection and RNA analysis Tissues were collectedand stored in RNAlater (Ambion, Austin, TX, USA), andRNA was extracted using the TRIzol Reagent (Invitrogen,Carlsbad, CA, USA). Samples of mRNA were analysedwith SYBR Green-based real-time quantitative RT-PCR(Applied Biosystems, Foster City, CA, USA), with β-actinor cyclophilin A as the reference gene in each reaction.Sequences of the primers used for RT-PCR assays areshown in Electronic supplementary material (ESM) Table 1.

Protein analysis For the insulin signalling, 62.5 mU/kginsulin was administered through the portal vein, andmuscle tissues were collected 5 min after injection asdescribed by Hirosumi et al. [27]. Liver and WAT tissuesamples were collected 2 min after injection with 200 and500 mU/kg insulin through the portal vein, respectively.Further methods can be found in the ESM.

Results

Upregulation of TLR2 and inflammatory cytokines in WATand liver of obese mice To examine the association ofinflammation with obesity and insulin resistance, wedetermined the gene expression of inflammatory mediatorsin the major insulin-responsive tissues of ob/ob mice. Wefound that mRNA expression of Tnf was significantlyelevated in WAT and liver, but not in muscle, of ob/obmice, compared with those of WT controls (Fig. 1a).Expression of Il1b and Il6 mRNA was increased in WAT,but not in liver and muscle, of ob/ob mice. The upregula-tion of inflammatory genes was accompanied by signifi-cantly increased expression of TLR2 and moderateinduction of TLR4 expression as shown in levels of bothmRNA and protein (Fig. 1b,c). Consistent with this, theupregulation of TLR2 was evident in WAT and liver, but

not in muscle. These results imply that obesity and nutrientexcess elicit upregulation in expression of inflammatorycytokines and TLR2 predominantly in WAT and liver. Thisprompted us to test directly whether TLR2 and its inducedinflammation are involved in metabolic disturbances.

Increased insulin sensitivity in Tlr2−/− mice To investigatethe role of TLR2 in glucose homeostasis, we examined theplasma levels of glucose and insulin in Tlr2−/− and WTmice. Plasma glucose and insulin levels after fasting wereboth significantly lower in Tlr2−/− mice than in those ofRC-fed WT mice (Fig. 2a,b). Consistent with thesefindings, Tlr2−/− mice fed an HF diet had significantlylower glucose and insulin levels relative to their WTcontrols. To assess the effect of TLR2 deficiency on thewhole-body glucose utilisation, we performed an OGTT.Tlr2−/− mice cleared glucose faster than WT mice,regardless of diet, indicating improved glucose tolerancein Tlr2−/− mice (Fig. 2c,d). This improved glucosetolerance was accompanied by significantly decreasedplasma insulin levels during OGTT of Tlr2−/− mice,regardless of diet (Fig. 2e,f). Thus, the insulin-resistanceindex calculated from the OGTT was significantly lower inTlr2−/− mice relative to their WT controls with either diet(Fig. 2g). Consistent with this, insulin tolerance tests (ITT)showed that glucose-lowering effects of insulin wereimproved in Tlr2−/− mice regardless of diet (Fig. 2h,i).These results demonstrated that the absence of TLR2increases glucose tolerance and insulin sensitivity regard-less of feeding mice RC or an HF diet.

Decreased body weight and fat mass in Tlr2−/− mice Wenext asked whether loss of TLR2 alters energy homeostasisin mice. The growth curve of Tlr2−/− mice was the same asthat of WT mice when mice were fed RC, but significantlydiffered in mice fed the HF diet (Fig. 3a,b). Bodycomposition analysis by echoMRI identified significantly

Fig. 1 Expression of inflammatory cytokines and TLRs in WAT,liver and gastrocnemius muscle of ob/ob mice. (a) Il1b, Il6 andTnf, and (b) Tlr2 and Tlr4 mRNA levels in tissues from male ob/obmice (n=6–8) relative to samples from WT mice (n=6–10). cProtein levels of TLR2 and TLR4 in tissues from male ob/ob andWT mice. Samples from representative animals are shown in the

western blot, with each lane representing one animal. Theintensities of the bands, quantified densitometrically relative toWT, are shown with sample number in parentheses. White bars,WT; grey bars, ob/ob. *p<0.05, **p<0.01 and ***p<0.001compared with WT mice

170 Diabetologia (2011) 54:168–179

reduced fat mass and increased lean mass in aged (11 monthold) Tlr2−/− mice, while no difference in fat and lean masswas found between young (2 month old) Tlr2−/− and WTmice (Fig. 3c). Furthermore, we found significant decreasesin individual fat mass of aged Tlr2−/− mice (Fig. 3d). Theintra-abdominal, including gonadal and retroperitoneal,WAT masses in Tlr2−/− mice were about 35% of those inWT mice. A similar reduction was also observed in thesubcutaneous inguinal WAT mass of Tlr2−/− mice, althoughthe difference did not reach statistical significance. Theinterscapular brown adipose tissue (BAT) weight of Tlr2−/−

mice was about 70% that of WT mice. The weights ofmajor organ, including kidney, liver, spleen and heart, wereindistinguishable between genotypes. Gastrocnemius skel-etal muscle weight of Tlr2−/− mice was significantlyincreased. These data indicate that the absence of TLR2causes reductions in fat deposition and content. While HFfeeding increased body weight of WT mice, the increases inbody weight of Tlr2−/− mice were significantly diminished(Fig. 3b). Consistent with this, the individual WAT andBAT masses were significantly less in Tlr2−/− mice fed anHF diet than in WT mice (Fig. 3e).

Microscopically, adipocytes in gonadal WAT of RC-fedaged Tlr2−/− mice were smaller than WT cells (mean area821 μm2 in Tlr2−/− vs 1503 μm2 in WT) (Fig. 3f,g). Thenumbers of adipocytes were similar in the Tlr2−/− and WTmice (1.89×107 in TLR2−/− vs 1.73×107 in WT). HFfeeding led to fat deposition in both Tlr2−/− and WT miceand the difference in adipocyte size between genotypes was

preserved (mean area 2242 μm2 in Tlr2−/− vs 3276 μm2 inWT) (Fig. 3h,i). The decrease in adipocyte size of Tlr2−/−

mice under HF feeding was associated with a decrease inthe number of adipocytes (1.21×107 in Tlr2−/− vs 2.62×107 in WT). Thus, the deficiency of TLR2 brought a shift inthe distribution of adipocyte sizes towards smaller adipo-cytes in WAT (Fig. 3g,i). The reduction in fat mass ofTlr2−/− mice appears to result primarily from the reductionin adipocyte size with a modest decrease in adipocytenumber. In vitro differentiation of embryonic fibroblastsshowed that adipogenic ability judged by the formation ofcells staining positive with Oil red O after hormonestimulation was preserved in Tlr2−/− mice (Fig. 3j), rulingout the possibility of defective adipogenic programmingwith TLR2 deficiency.

The critical variables contributing to body weightmaintenance include energy intake and expenditure. Meta-bolic analysis showed that Tlr2−/− mice had similar dailyfood intake to WT mice (Fig. 4a), suggesting that Tlr2−/−

mice have normal energy absorption. Body temperaturewas modestly increased in male and significantly increasedin female Tlr2−/− mice, regardless of diet (Fig. 4b). Basalenergy expenditure, as measured by oxygen consumption(IVO2), during the dark phase was dramatically increased inTlr2−/− mice (Fig. 4c,d). As the increases in

IVO2 and

IVCO2 were coordinated, the respiratory exchange ratio wasunchanged (Fig. 4e,f). Consistent with this, vertical rearingof both young and aged Tlr2−/− mice during the dark phasewas markedly increased (Fig. 4g). Ambulatory locomotor

Fig. 2 Glucose homeostasis andinsulin sensitivity in Tlr2−/−

mice. a Fasting plasma glucosein male mice fed RC and HFdiets (WT RC, n=15; Tlr2−/−

RC, n=16; WT HF, n=16;Tlr2−/− HF, n=12). b Fastinginsulin levels in male mice fedRC and HF diets (WT RC, n=8;Tlr2−/− RC, n=7; WT HF, n=14; Tlr2−/− HF, n=9). (c,d)Plasma glucose during OGTT inmale mice fed (c) RC and (d)HF diets. (e,f) Insulin levelsduring OGTT in male mice fed(e) RC and (f) HF diets. gInsulin-resistance (IR) index dur-ing OGTT in male mice fed RCand HF diets. (h,i) ITT results formale mice fed (h) RC and (i) HFdiets. White bars/symbols, WT;black bars/symbols, Tlr2−/−. *p<0.05, **p<0.01 and ***p<0.001compared with WT mice

Diabetologia (2011) 54:168–179 171

activity of aged Tlr2−/− mice during both the dark and lightphases was significantly increased (Fig. 4h). Thus, energydissipation appeared increased in Tlr2−/− mice.

Lipid profile and inflammatory mediators in Tlr2−/−

mice WAT actively secretes signalling molecules, includingNEFA, adipokines and inflammatory molecules, into thecirculation and communicates with other organs to regulateinsulin sensitivity [28]. Fasting plasma triacylglycerol andNEFA levels were not distinguishable between genotypesin mice fed RC, and were significantly lower in Tlr2−/−

mice than in WT mice fed an HF diet (Fig. 5a,b). Plasmacholesterol levels were not different between genotypes,regardless of diet (Fig. 5c). Circulating leptin and resistin

levels were not distinguishable between genotypes in micefed RC, but were reduced in Tlr2−/− mice fed an HF diet(Fig. 5d,e). Adiponectin levels in circulation were compa-rable between Tlr2−/− and WT mice, regardless of diet(Fig. 5f). While circulating inflammatory molecules, in-cluding IL-6, monocyte chemotactic protein-1 (MCP-1) andTNF-α, were higher in mice fed an HF diet than in thosefed RC, no difference was detectable between genotypes,regardless of diet (Fig. 5g–i). Although the decreased levelsof plasma triacylglycerol, NEFA and resistin may accountfor increased insulin sensitivity in Tlr2−/− mice upon HFfeeding, the increased insulin sensitivity was, however,observed in Tlr2−/− mice fed RC despite any significantchange in plasma triacylglycerol, NEFA and resistin. These

Fig. 3 Body composition of Tlr2−/− mice. (a,b) Body weights offemale mice fed (a) RC (WT, n=8; Tlr2−/−, n=7) and (b) HF diets(WT, n=7; Tlr2−/−, n=7); at first time point after HF was started, p<0.05; at time point 2, p<0.01; at all subsequent time points, p<0.001.c Body composition analyses of 2 and 11 month old male mice byMRI (2 month old mice, n=5; WT 11 month old mice, n=7; Tlr2−/−

11 month old mice, n=9). (d,e) Organ weights in (d) 11 month oldfemale WT (n=8) and Tlr2−/− (n=7) mice fed RC and (e) 7 month oldfemale WT (n=7) and Tlr2−/− (n=7) mice fed the HF diet. Gon,

gonadal WAT; Ing, inguinal WAT; Ret, retroperitoneal WAT. Muscleindicates gastrocnemius skeletal muscle. (f,h) Morphology and (g,i)cell size distribution of gonadal WAT from WT and Tlr2−/− mice fed(f,g) RC and (h,i) HF diets. The scale bar indicates 100 μm for allimages. j In vitro adipogenesis in embryonic fibroblasts. Oil red Ostaining of plates after 8 days of differentiation. White bars/symbols,WT; black bars/symbols, Tlr2−/−. *p<0.05, **p<0.01 and ***p<0.001 compared with WT mice

172 Diabetologia (2011) 54:168–179

results suggest that the deficiency of TLR2 has limitedeffects on systemic levels of inflammatory mediators.

Attenuated local inflammatory cytokine expression andsignalling in Tlr2−/− mice Although the deficiency ofTLR2 did not elicit apparent differences in the systemic levelsof inflammatory molecules, insulin sensitivity can be affectedthrough the change in local insulin-responsive tissues. Wenext examined whether the absence of TLR2 would attenuatelocal inflammation. Tlr2−/− RC-fed mice exhibited, in liver, asignificant reduction in expression of Il6 and a tendencytowards decreased Il1b expression, but no change inexpression of Tnf and macrophage marker Emr1 (Fig. 6a).No difference in expression of these cytokines was identifiedin WAT and muscle of Tlr2−/− mice. Interestingly, dramaticreductions in expression of Il1b, Il6 and Emr1 as well as atendency towards a decrease in expression of Tnf wereobserved in WAT and liver of Tlr2−/− mice fed the HF diet(Fig. 6b). Expression of these genes, except Tnf, remainedunaltered in muscle of Tlr2−/− HF-fed mice (Fig. 6b).Because TLR2 has been demonstrated to transduce signal toactivate protein kinases, including IKK, JNK and ERK [7–9], we examined their phosphorylation in the insulin-responsive tissues. No difference in phosphorylation ofIKKα/β and JNK between WT and Tlr2−/− mice wasdetectable in WAT, liver and muscle, regardless of diet(Fig. 6c,d). While phosphorylation of ERK1/2 was similar inWAT and muscle in the WT and Tlr2−/− mice fed RC, its

phosphorylation was markedly decreased in the liver ofTlr2−/− mice (Fig. 6c). The decrease in hepatic ERK1/2phosphorylation was not accompanied by the change ofERK1/2 protein content in the liver. Furthermore, the liver-specific decrease in ERK1/2 phosphorylation was alsoobserved in Tlr2−/− mice fed the HF diet (Fig. 6d). Thus,the absence of TLR2 attenuates basal inflammatory cytokineexpression and signalling specifically in the liver.

Inflammation-induced activation of IKK, JNK and ERKhas been shown to phosphorylate IRS1 at its serine residue(s), which further interrupts tyrosine phosphorylation onIRS1 and suppresses insulin signalling [7–9]. To furtherevaluate whether the attenuated inflammatory cytokineexpression and signalling in the liver of Tlr2−/− micecorrelates with serine phosphorylation of IRS1, we detectedthe phosphorylation at Ser307 and Ser612 of IRS1. Nodifference in phosphorylation at Ser307 and Ser612 of IRS1between WT and Tlr2−/− livers was detected, regardless ofdiet (Fig. 6e,f). These results suggest that TLR2 deficiencydid not affect phosphorylation at Ser307 or Ser612 of IRS1in the liver.

Improved insulin signalling specifically in the liver ofTlr2−/− mice To determine which insulin-responsive tissue(s)exhibited improved insulin signalling in Tlr2−/− mice, weexamined the insulin action in mice fed RC. Insulin-stimulated phosphorylation at Ser473 of protein kinase B(Akt) was increased in liver, but not in WAT and muscle of

Fig. 4 Energy homeostasis in Tlr2−/− mice. a Daily food intake inmetabolism cages (n=5 for all results). b Body temperature measured at10:00 hours in WT and Tlr2−/− mice (RC male [M] mice, n=8; WT RCfemale [F] mice, n=8; Tlr2−/− F mice, n=7; WT HF M mice, n=8;Tlr2−/− M mice, n=5; WT HF F mice, n=7; Tlr2−/− F mice, n=7. (c)Oxygen consumption (

IVO2) and (d) mean AUC; (e) respiratoryexchange ratio (RER) and (f) mean 12 h values, of 2 month old malemice in indirect calorimetry over 48 h (WT, n=5; Tlr2−/−, n=4). Bold

bars on the x-axes represent the dark phases. (g, h) Locomotor activitymeasured as (g) the vertical rearing and (h) the ambulatory activity,across a 30 min period of observation (2 month old mice, n=5; WT11 month old mice, n=7; Tlr2−/− 11 month old mice, n=9; D, dark;L, light). White bars/symbols, WT; black bars/symbols, Tlr2−/−. *p<0.05, **p<0.01,***p<0.001 and †p=0.0695 compared with WTmice

Diabetologia (2011) 54:168–179 173

Tlr2−/− mice, compared with those of WT mice (Fig. 7a).This was accompanied by increased tyrosine phosphoryla-tion of IRS1 and association of IRS1: phosphoinositide 3-kinase (PI3K) complex in the livers of Tlr2−/− mice afterinsulin stimulation. However, the increases in IRS1 tyrosinephosphorylation and IRS1:PI3K association were not foundin WAT and muscle of Tlr2−/− mice. Consistently, the liver-specific increases in insulin-stimulated Akt phosphorylation,IRS1 tyrosine phosphorylation and IRS1:PI3K association(data not shown) were also observed in Tlr2−/− mice fed HFdiet (Fig. 7b). Thus, liver appears to be the major tissue withincreased insulin signalling in Tlr2−/− mice.

Improved glucose metabolism in the liver of Tlr2−/−

mice To determine the functional correlates of increasedinsulin signalling in the livers of Tlr2−/− mice, wedetermined aspects of hepatic glucose metabolism. Pyru-vate tolerance tests showed that glucose levels weresignificantly lower in Tlr2−/− mice in response to theadministration of pyruvate, a major gluconeogenesis sub-strate (Fig. 8a). We next assessed the effect of TLR2deficiency in primary hepatocytes isolated from Tlr2−/− and

WT mice. TLR2 deficiency significantly reduced pyruvate-induced glucose release into the medium by primaryhepatocytes (Fig. 8b). To examine the functional effects ofTLR2 deficiency on hepatocyte insulin responsiveness, theability of insulin to inhibit hepatocyte glucose productionwas determined. Glucose production by WT hepatocyteswas reduced by ∼20% within this time period in response toinsulin (Fig. 8c). The inhibition of glucose secretion byinsulin in Tlr2−/− hepatocytes was significantly increasedby 50% in comparison with WT hepatocytes. Thus,increased insulin signalling secondary to TLR2 deficiencyresulted in an enhancement of insulin to suppress hepato-cyte glucose secretion. Together, these results indicateTlr2−/− mice have reduced hepatic glucose production andincreased response to insulin.

To determine the insulin sensitivity in the skeletalmuscle, we performed the in vitro glucose uptake assay.Basal glucose uptake in muscle in vitro was similarbetween Tlr2−/− and WT mice (Fig. 8d). Insulin-stimulated glucose uptake in muscle of Tlr2−/− mice was1.6 times those of WT, although the difference did notreach the statistical significance. Consistent with this,triacylglycerol content in liver, which is associated withhepatic insulin resistance, was significantly reduced inTlr2−/− mice, while muscle triacylglycerol content wasnot altered in Tlr2−/− mice (Fig. 8e).

Discussion

Substantial evidence indicates inflammation as a keymechanism linking metabolic disturbance to nutrient ex-cess. To establish this link, we hypothesised that TLR2transduces the inflammatory signals and further contributesto insulin resistance. Our findings recapitulated the obser-vation that obesity is associated with elevated expression ofinflammatory cytokines in WAT and liver. Moreover, micelacking TLR2 exhibited enhanced insulin sensitivity, whichis accompanied by increased transduction of insulin signal,improved glucose metabolism, and attenuated inflammatorycytokine expression and related signalling specifically inthe liver. Thus, our work has identified TLR2 as a keymediator of hepatic inflammation-related signalling andinsulin resistance.

Body weight represents the homeostasis betweenenergy intake and expenditure. The deficiency of TLR2caused a reduction in body weight and fat mass withoutalteration in food intake. Increased energy expenditureand locomotor activity, as well as body temperature,point to increased energy dissipation in Tlr2−/− mice. It isknown that increased physical activity can influence theinsulin sensitivity of skeletal muscle, which is consistentwith our findings of a modest increase of insulin-

Fig. 5 Lipid profiles and inflammatory mediators in Tlr2−/− mice.Plasma levels of (a) triacylglycerol, (b) NEFA, (c) cholesterol, (d)leptin, (e) resistin, (f) adiponectin, (g) IL-6, (h) MCP-1 and (i) TNF-αin male mice fed RC and HF diets (WT, n=8; Tlr2−/−, n=5). Whitebars, WT; black bars, Tlr2−/−. *p<0.05 and **p<0.01 for Tlr2−/−

mice compared with WT mice

174 Diabetologia (2011) 54:168–179

stimulated glucose uptake in Tlr2−/− muscle. However, thedifferences in body weight and fat mass were notnoticeable unless mice were aged up to 11 months orstimulated with a high-energy diet. The reductions in WATmass and adipocyte size of Tlr2−/− mice could result fromthe defect of pre-adipocytes differentiating into adipocytesand/or maturation of adipocytes. To address the bona fiderole of TLR2 in adipocyte differentiation and maturation,we subjected embryonic fibroblasts to adipogenesis invitro. TLR2 deficiency did not influence the abilities of

adipocyte differentiation and triacylglycerol accumulation,suggesting that these properties are not affected by TLR2deficiency. Although the mechanism involved in increasedenergy expenditure in Tlr2−/− mice is not clear, thereduced deposition of excess energy in the body poten-tially contributes to better metabolic profiles and glucosemetabolism.

Improvement of insulin sensitivity with TLR2 defi-ciency is present in mice fed RC, suggesting TLR2-induced insulin resistance takes place in the basal

Fig. 6 Local inflammatory cytokine expression and signalling in Tlr2−/−

mice. Il1b, Il6, Tnf and Emr1 mRNA levels in Tlr2−/− (n=5–7) tissuesrelative to WT (n=7–8) of mice fed (a) RC and (b) HF diets. Whitebars, WT; black bars, Tlr2−/−. Immunoblot analyses on phosphorylationof IKKα/β, JNK and ERK1/2 and protein content of ERK1/2 ingonadal WAT, liver and gastrocnemius muscle from WT and Tlr2−/−

mice fed (c) RC and (d) HF diets. Immunoblot analyses on

phosphorylation at Ser307 and Ser612 of IRS1 in the liver of WT andTlr2−/− mice on (e) RC and (f) HF diets. Each band represents a tissueextract from a single mouse. The intensities of the bands, quantifieddensitometrically relative to WT, are shown with the sample number inparentheses. *p<0.05, **p<0.01 and ***p<0.001 for Tlr2−/− micecompared with WT mice

Diabetologia (2011) 54:168–179 175

physiological condition without the need of further HFstimulation. This raises the speculation that TLR2 ligandand the signalling it mediates are present in the basalstate. Although the actual endogenous TLR2 ligandshave not been identified conclusively, it has beendemonstrated that TLR2 recognises a large number oflipid-containing molecules [18], as well as endogenousproteins released from damaged tissues or injured cells

[29]. It is interesting to note that the critical portion ofmany TLR2 ligands contains a similar fatty acid compo-nent. Removal or change of these fatty acids results in lossof TLR2 activation capability [30], implicating an impor-tant role of these fatty acids in ligand recognition andreceptor activation. Particularly, palmitic and stearic acidsare known to be major fatty acids acylated in thelipoprotein or lipopeptide to activate TLR2 [31, 32].

Fig. 8 Glucose metabolism and triacylglycerol content in liver andmuscle of Tlr2−/− mice. a Plasma glucose levels during PTT in malemice fed RC (n=4 each). b Basal and pyruvate-induced glucoserelease into the medium after 4 h incubation in primary hepatocytesisolated from Tlr2−/− and WT mice. c Insulin-suppressed glucoserelease into the medium after 16 h of incubation in primaryhepatocytes isolated from Tlr2−/− and WT mice. Results are expressed

as the percentage inhibition of glucose secretion. d Basal and insulin-stimulated glucose uptake in isolated gastrocnemius muscle fromTlr2−/− and WT mice. Results are derived from two independentexperiments performed in triplicate. e Triacylglycerol (TG) content inliver and gastrocnemius muscle of WT (n=8) and Tlr2−/− (n=10)mice. White bars/symbols, WT; black bars/symbols, Tlr2−/−. *p<0.05,**p<0.01 and ***p<0.001

Fig. 7 Insulin signalling inTlr2−/− mice. Immunoblot anal-yses on phosphorylation atSer473 of Akt, and detection ofIRS1 phosphotyrosine level andIRS1:PI3K association by IRS1immunoprecipitation in mice fed(a) RC and (b) HF diets. Eachband represents a tissue extractfrom a single mouse. The inten-sities of the bands relative toWT after insulin stimulation areshown with the sample numberin parentheses. Ins, insulin. *p<0.05 and **p<0.01 for Tlr2−/−

mice compared with WT mice

176 Diabetologia (2011) 54:168–179

NEFA possess the capacity to induce stress/inflammatorysignals not only in immune cells but also other cell typesand tissues [13, 14, 33]. This phenomenon is of particularimportance in conditions of nutrient excess such as obesityand diabetes or after ingestion of a fatty meal.

Inflammation-induced activation of protein kinases, suchas IKK, JNK and ERK, has been shown to inhibit tyrosinephosphorylation of IRS1 [7–9] and suppress insulin actionand its downstream signalling. Serine phosphorylation ofIRS1 is a general mechanism to interrupt IRS1 function andinsulin-signal transduction [34, 35]. For example, phos-phorylation of IRS1 at Ser307 reduces IRS1 coupling toactivated insulin receptors [8] and enhances IRS1 degrada-tion [36]. Phosphorylation of IRS1 at Ser612 decreasedPI3K docking to IRS1 [37]. IRS1 Ser307 can be phosphor-ylated by IKK and JNK [7, 8], whereas IRS1 Ser612 can bephosphorylated by ERK [9]. Our results showing enhancedinsulin-stimulated Akt Ser473 phosphorylation and IRS1tyrosine phosphorylation specifically in the liver of Tlr2−/−

mice provide a biochemical correlate for increased hepaticinsulin sensitivity. In the search for protein kinases involvedin insulin signalling that are induced by inflammatorymediators, we did not detect alterations in the activation ofJNK and IKK in the livers of Tlr2−/− mice. ERK activationwas dramatically reduced in the livers of Tlr2−/− mice,providing a possible link between TLR2 and hepaticinflammation. However, further investigation of the hypoth-esised phosphorylation target on IRS1 did not reveal anychange in phosphorylation of Ser307 and Ser612. Theseresults suggest that the hypothesised phosphorylation targeton IRS1 mediated through TLR2 is not Ser307 or Ser612.Nevertheless, our results implicate ERK as a downstreamintracellular mediator of the TLR2-induced inflammatoryresponse influencing the insulin signal pathway throughIRS1 modification.

Our observation of improved insulin sensitivity in theabsence of TLR2 supports and elaborates previousfindings. For example, palmitate induced IL-6 productionand inflammatory signalling, leading to inhibition ofinsulin-activated signal transduction through TLR2 inmyotubes [21]. In addition, inhibition of TLR2 expressionby TLR2 antisense oligonucleotide improves insulinsensitivity and signalling in muscle and WAT from micefed the HF diet [38]. However, tissue specificity andtransient treatment of antisense oligonucleotide limit theinterpretation of long-term effect of TLR2 on complexphysiological homeostasis. Although mice lacking TLR2have recently been found to be protected from diet-induced adiposity and systemic insulin resistance [39],mechanistic insight and tissue specificity for the improve-ment of insulin sensitivity with TLR2 deficiency havebeen lacking. Collectively, these findings implicate TLR2as a key mediator of inflammation, causing insulin

resistance under conditions of nutrient excess in manykey insulin-responsive tissues.

Despite the apparent upregulation of inflammatorycytokines in WAT of ob/ob mice, the induction ofinflammatory cytokines and TLR2 was also present in thelivers of ob/ob mice in our study. These findings suggestthat the nutrient-induced inflammatory response and insulinresistance takes place in both WAT and liver. In support ofthis hypothesis, TLR2-deficiency-associated decreases ininflammatory cytokine expression and signalling andenhancement of insulin signalling were conspicuous in theliver. In contrast, the induction of inflammatory cytokineswas not evident in the muscle of ob/ob mice. Consistentwith this, the gene expression and signalling moleculesrelated to inflammation and insulin resistance in muscleappeared unchanged in Tlr2−/− mice. Thus, the modifica-tion of peripheral insulin sensitivity through TLR2-inducedinflammatory cytokine expression and signalling is likely tobe mediated primarily in the liver.

Recently, the role of TLR4 in the cross-talk betweennutrient-induced inflammation and insulin resistance hasbeen investigated [40–43], and the hypothesised locationfor TLR4 action mediating the impairment of insulinsensitivity is predominantly in WAT. Different TLRproduction profiles affect the responsiveness of particularcell types and tissues to the ligand stimulation. Inaddition, different TLR ligands may induce activationof different downstream signalling pathways, leading to adiverse array of target gene expression and cellularresponses through distinct TLRs. Given the differencesin ligand preference, signalling pathways and tissuedistribution between TLR2 and TLR4, it is reasonableto speculate that TLR2 and TLR4 may receive differentligand stimulation from nutrient factors and transduceinflammatory signals. Furthermore, there was no com-pensatory alteration in the gene expression of Tlr4 inWAT, liver and muscle (data not shown). Our resultsunequivocally provide TLR2 as another TLR familyreceptor involved in the interface of inflammatoryresponse and metabolic disturbance.

In conclusion, our results demonstrated that TLR2contributes to upregulation of inflammatory cytokineexpression and signalling and interference of insulin actionspecifically in the liver. Although the detailed mechanismby which factors activate TLR2 or interact with itsassociated factors is not known, the results presented hererepresent a new paradigm that endogenous molecule(s) canutilise the innate immune receptor TLR2 to trigger the localinflammation-related signalling in the liver and subsequent-ly affect systemic metabolism. Our findings provide arationale for the development of the hepatic inflammatoryresponse and insulin resistance in response to nutrientfactors.

Diabetologia (2011) 54:168–179 177

Acknowledgements We thank A. Pendse, N. Takahashi and C.-H.Lee for discussions; Z.-H. Lin, H.-F. Jheng, the National LaboratoryAnimal Centre Pathology Core (C.-T. Liang), H.-T. Wu, Y.-H. Lee andthe Taiwan Mouse Clinic for technical assistance; and S. Akira forkindly providing Tlr2−/− mice. This work was supported by the grantfrom the National Health Research Institute (NHRI-EX98-9823SC).

Duality of interest The authors declare that there is no duality ofinterest associated with this manuscript.

References

1. Wellen KE, Hotamisligil GS (2005) Inflammation, stress, anddiabetes. J Clin Invest 115:1111–1119

2. Shoelson SE, Lee J, Goldfine AB (2006) Inflammation andinsulin resistance. J Clin Invest 116:1793–1801

3. Kim JK, Kim YJ, Fillmore JJ et al (2001) Prevention of fat-inducedinsulin resistance by salicylate. J Clin Invest 108:437–446

4. Cai D, Yuan M, Frantz DF et al (2005) Local and systemic insulinresistance resulting from hepatic activation of IKK-beta and NF-kappaB. Nat Med 11:183–190

5. Shoelson SE, Lee J, Yuan M (2003) Inflammation and the IKKbeta/I kappa B/NF-kappa B axis in obesity- and diet-inducedinsulin resistance. Int J Obes Relat Metab Disord 27(Suppl 3):S49–S52

6. Yuan M, Konstantopoulos N, Lee J et al (2001) Reversal ofobesity- and diet-induced insulin resistance with salicylates ortargeted disruption of Ikkβ. Science 293:1673–1677

7. Gao Z, Hwang D, Bataille F et al (2002) Serine phosphorylationof insulin receptor substrate 1 by inhibitor kappa B kinasecomplex. J Biol Chem 277:48115–48121

8. Aguirre V, Uchida T, Yenush L, Davis R, White MF (2000) The c-Jun NH(2)-terminal kinase promotes insulin resistance duringassociation with insulin receptor substrate-1 and phosphorylationof Ser(307). J Biol Chem 275:9047–9054

9. De Fea K, Roth RA (1997) Modulation of insulin receptorsubstrate-1 tyrosine phosphorylation and function by mitogen-activated protein kinase. J Biol Chem 272:31400–31406

10. Stamler R, Stamler J, Riedlinger WF, Algera G, Roberts RH(1978) Weight and blood pressure. Findings in hypertensionscreening of 1 million Americans. JAMA 240:1607–1610

11. Bonora E, Kiechl S, Willeit J et al (1998) Prevalence of insulinresistance in metabolic disorders: the Bruneck Study. Diabetes47:1643–1649

12. Kahn BB, Flier JS (2000) Obesity and insulin resistance. J ClinInvest 106:473–481

13. Ghanim H, Aljada A, Hofmeyer D, Syed T, Mohanty P, DandonaP (2004) Circulating mononuclear cells in the obese are in aproinflammatory state. Circulation 110:1564–1571

14. Jove M, Planavila A, Laguna JC, Vazquez-Carrera M (2005)Palmitate-induced interleukin 6 production is mediated by proteinkinase C and nuclear-factor kappaB activation and leads toglucose transporter 4 down-regulation in skeletal muscle cells.Endocrinology 146:3087–3095

15. Tschop M, Thomas G (2006) Fat fuels insulin resistance throughToll-like receptors. Nat Med 12:1359–1361

16. Akira S, Takeda K, Kaisho T (2001) Toll-like receptors: criticalproteins linking innate and acquired immunity. Nat Immunol2:675–680

17. Medzhitov R (2001) Toll-like receptors and innate immunity. NatRev Immunol 1:135–145

18. Akira S (2003) Mammalian toll-like receptors. Curr OpinImmunol 15:5–11

19. Buwitt-Beckmann U, Heine H, Wiesmuller KH, Jung G, Brock R,Ulmer AJ (2005) Lipopeptide structure determines TLR2 depen-dent cell activation level. FEBS J 272:6354–6364

20. Buwitt-Beckmann U, Heine H, Wiesmuller KH et al (2006)TLR1- and TLR6-independent recognition of bacterial lipopep-tides. J Biol Chem 281:9049–9057

21. Senn JJ (2006) Toll-like receptor-2 is essential for the develop-ment of palmitate-induced insulin resistance in myotubes. J BiolChem 281:26865–26875

22. Murakami K, Bujo H, Unoki H, Saito Y (2007) High fat intakeinduces a population of adipocytes to co-express TLR2 andTNFalpha in mice with insulin resistance. Biochem Biophys ResCommun 354:727–734

23. NguyenMT, Favelyukis S, Nguyen AK et al (2007) A subpopulationof macrophages infiltrates hypertrophic adipose tissue and isactivated by free fatty acids via Toll-like receptors 2 and 4 andJNK-dependent pathways. J Biol Chem 282:35279–35292

24. Lin Y, Lee H, Berg AH, Lisanti MP, Shapiro L, Scherer PE (2000)The lipopolysaccharide-activated toll-like receptor (TLR)-4 indu-ces synthesis of the closely related receptor TLR-2 in adipocytes.J Biol Chem 275:24255–24263

25. Lang CH, Silvis C, Deshpande N, Nystrom G, Frost RA (2003)Endotoxin stimulates in vivo expression of inflammatory cyto-kines tumor necrosis factor alpha, interleukin-1beta, -6, and high-mobility-group protein-1 in skeletal muscle. Shock 19:538–546

26. Takeuchi O, Hoshino K, Kawai T et al (1999) Differential roles ofTLR2 and TLR4 in recognition of gram-negative and gram-positive bacterial cell wall components. Immunity 11:443–451

27. Hirosumi J, Tuncman G, Chang L et al (2002) A central role forJNK in obesity and insulin resistance. Nature 420:333–336

28. Rosen ED, Spiegelman BM (2006) Adipocytes as regulators ofenergy balance and glucose homeostasis. Nature 444:847–853

29. Rubartelli A, Lotze MT (2007) Inside, outside, upside down:damage-associated molecular-pattern molecules (DAMPs) andredox. Trends Immunol 28:429–436

30. Brightbill HD, Libraty DH, Krutzik SR et al (1999) Host defensemechanisms triggered by microbial lipoproteins through toll-likereceptors. Science 285:732–736

31. Fujita M, Into T, Yasuda M et al (2003) Involvement of leucineresidues at positions 107, 112, and 115 in a leucine-rich repeatmotif of human Toll-like receptor 2 in the recognition ofdiacylated lipoproteins and lipopeptides and Staphylococcusaureus peptidoglycans. J Immunol 171:3675–3683

32. Muhlradt PF, Kiess M, Meyer H, Sussmuth R, Jung G (1997)Isolation, structure elucidation, and synthesis of a macrophagestimulatory lipopeptide from Mycoplasma fermentans acting atpicomolar concentration. J Exp Med 185:1951–1958

33. Nguyen MT, Satoh H, Favelyukis S et al (2005) JNK and tumornecrosis factor-alpha mediate free fatty acid-induced insulinresistance in 3 T3-L1 adipocytes. J Biol Chem 280:35361–35371

34. Gual P, Le Marchand-Brustel Y, Tanti JF (2005) Positive andnegative regulation of insulin signaling through IRS-1 phosphor-ylation. Biochimie 87:99–109

35. Taniguchi CM, Emanuelli B, Kahn CR (2006) Critical nodes insignalling pathways: insights into insulin action. Nat Rev MolCell Biol 7:85–96

36. Greene MW, Sakaue H, Wang L, Alessi DR, Roth RA (2003)Modulation of insulin-stimulated degradation of human insulinreceptor substrate-1 by Serine 312 phosphorylation. J Biol Chem278:8199–8211

37. Mothe I, Van Obberghen E (1996) Phosphorylation of insulinreceptor substrate-1 on multiple serine residues, 612, 632, 662,and 731, modulates insulin action. J Biol Chem 271:11222–11227

38. Caricilli AM, Nascimento PH, Pauli JR et al (2008) Inhibition oftoll-like receptor 2 expression improves insulin sensitivity and

178 Diabetologia (2011) 54:168–179

signaling in muscle and white adipose tissue of mice fed a high-fatdiet. J Endocrinol 199:399–406

39. Himes RW, Smith CW (2010) Tlr2 is critical for diet-inducedmetabolic syndrome in a murine model. FASEB J 24:731–739

40. Shi H, Kokoeva MV, Inouye K, Tzameli I, Yin H, Flier JS (2006)TLR4 links innate immunity and fatty acid-induced insulinresistance. J Clin Invest 116:3015–3025

41. Tsukumo DM, Carvalho-Filho MA, Carvalheira JB et al(2007) Loss-of-function mutation in Toll-like receptor 4

prevents diet-induced obesity and insulin resistance. Diabetes56:1986–1998

42. Poggi M, Bastelica D, Gual P et al (2007) C3H/HeJ mice carryinga toll-like receptor 4 mutation are protected against the develop-ment of insulin resistance in white adipose tissue in response to ahigh-fat diet. Diabetologia 50:1267–1276

43. Kim F, Pham M, Luttrell I et al (2007) Toll-like receptor-4mediates vascular inflammation and insulin resistance in diet-induced obesity. Circ Res 100:1589–1596

Diabetologia (2011) 54:168–179 179